Diabetic Retinopathy Management Guidelines

Rahul Chakrabarti; C Alex Harper; Jill Elizabeth Keeffe


Expert Rev Ophthalmol. 2012;7(5):417-439. 

In This Article

Assessment of DR: Who can Examine?

The availability of sufficient numbers of ophthalmologists to meet the growing demands around the world, particularly in developing regions, has emerged as a major barrier to delivery of timely ophthalmic care. Recent evidence suggests the problem is masked by inequities in distribution of the health workforce, whereby poor working and living conditions and greater income-earning capacity in urban areas means that medical staff are often reluctant to relocate to work in remote areas, and less willing to work in the government health system.[119,120] In order to implement sustainable guidelines, an approach that has been proposed was to 'task-shift' to increase reliance on community level and nonophthalmic workers in the process of DR screening.[121] Consensus was achieved among the guidelines that in addition to ophthalmologists, screening could be reliably performed by adequately trained doctors, retinal photographers and optometrists (Table 5). This was particularly emphasized in guidelines from developing regions including India and South Africa where the issue of adequate healthcare personnel has demanded innovative methods of delivering care. The SIGN and AAO guidelines, while recommending that ophthalmologists perform most of the examinations and surgery, acknowledged that 'trained individuals' could be involved in the screening process in order to improve access to care. Several studies comparing accuracy of other health professionals in detection and grading DR have been covered well in the literature.

The sensitivity of detecting vision-threatening retinopathy using direct ophthalmoscopy ranged from 41 to 87% among general practitioners;[122,123] 74–100% by optometrists/opticians[124,125] and 14–55% by nurses.[107,126] Buxton et al. compared the sensitivity of detecting vision-threatening retinopathy by hospital physicians and general practitioners in 3318 patients with DR using direct ophthalmoscopy in the UK.[122] General practitioners demonstrated a sensitivity and specificity of 41 and 89%; compared with 67 and 96%, respectively, for hospital physicians. Recently, Gill et al. evaluated the ability of 11 general practitioners to assess for referrable DR in 28 patients using a non-mydriatic panoptic ophthalmoscope.[123] The authors compared findings with a series of reference standard retinal diagrams. The results demonstrated a sensitivity of 87% with specificity of 57% for detecting referable DR. Despite these findings, a survey of DR screening practices by Australian family physicians found only 26% routinely examined their patients with DM for DR. The low rate for ophthalmoscopy was largely accounted for by the deficiency in confidence in detecting changes as reported by 84% of doctors surveyed.[127] Importantly, in the study by Gill et al., prior to examination general practitioners were required to participate in a 4-h tutorial program conducted by a retinal specialist. These findings were consistent with further studies that have demonstrated that the level of knowledge, and clinical skills for detection of DR increased after appropriate and standardized training.[128,129]

Several studies demonstrated that optometrists had a high sensitivity for the detection of retinopathy. Kleinstein et al., assessing the accuracy of optometrists using direct ophthalmoscopy in the UK, showed a sensitivity of 74% and specificity of 84% for the presence of DR.[124] Furthermore, the accuracy for diagnosis of retinopathy severity was comparable with general ophthalmologists. Importantly, Burnett et al., in a sample of patients referred from general practices in north London (UK), demonstrated that optometrists were able to assess referrable DR with 100% sensitivity and 94% specificity.[125] In addition, Schmid et al. conducted a comprehensive study of optometrist DR screening practices in northern Australia.[130] They demonstrated a combined approach integrating education of optometrists yielded an agreement of 79% with retinal specialists for appropriately identifying patients requiring specialist-level care.

The utilization of nurses and physician assistants for DR screening has also been explored with varying results. Pugh et al. demonstrated that dilated ophthalmoscopy conducted by trained physician assistants yielded a sensitivity of only 14%, with a specificity of 99%, for assessment of different severity levels of DR in 250 patients compared with the 'gold standard' ETDRS.[107] Furthermore, in a community-based setting, Forrest et al. showed that while the accuracy of nurses (sensitivity 50% and specificity 99%) was comparable with diabetologists for the detection of DR using dilated ophthalmoscopy, the ability to detect serious retinopathy was lower by nurses.[126]

Due to the variable accuracy of non-ophthalmic personnel to detect DR using ophthalmoscopy, evaluation of clinicians to read retinal images has also been evaluated. Farley et al. evaluated and assessed the accuracy of general practitioners (family physicians) compared with ophthalmologists to grade and appropriately refer retinal images taken using a single-field 45° non-mydriatic retinal camera, of 1040 predominantly Hispanic-background patients attending a general medical clinic. The authors concluded as a primary end point that general practitioners failed to refer only 10.2% of cases which ophthalmologists would have considered necessary.[131] Furthermore, the use of trained graders examining non-mydriatic images for detecting sight-threatening and referable DR has demonstrated a sensitivity of 85–97% and specificity of 80–96%.[107,132] In Spain, Andonegui et al. compared the accuracy of primary care physicians to ophthalmologist in reviewing five-field non-mydriatic photographs in a randomized sample of 200 patients, half with DR.[133] Primary care physicians received online clinical education prior to reviewing the images. The study showed agreement between physicians and ophthalmologists of between 80 and 95%. However, the study failed to assess accuracy in DR severity, which would be important for guiding referral. Nevertheless, a growing body of evidence suggests that dilated examination and reliable interpretation of non-mydriatic retinal photography can be performed by trained personnel to meet screening sensitivity criteria.